Effects Of Landscape Pattern On Runoff And Sediment In The Loess Plateau:A Multi-scale Study

Since the 1950s,a series of measures were implemented in the Loess Plateau to reduce the sediment discharge and maintain a healthy Yellow River.Consequently,dramatic changes in landscape pattern have been observed in the Loess Plateau,which inevitably affects hydrology process.A better understanding of the changes in river runoff and sediment loads over time and the driving mechanisms are thus of paramount importance for decision makers and planners to take appropriate sustainability measures,which also have important scientific significance.However,the effect of the landscape pattern on surface runoff and sediment yield at different scales remain uncertain.Thus,for slope scale,artificial simulated rainfall experiment were applied to evaluate the effect of landscape vegetation coverage and vegetation patches on flow resistance relations and sediment transport.Exploring overland flow resistance and sediment delivery capacity under different landscape vegetation conditions is critical for better understanding hydrodynamic mechanisms of soil erosion processes and hydrological connectivity on hillslopes.For watershed scale,the Soil and Water Assessment Tool（SWAT）model,RUSLE,double mass curve were used to investigate the contributions of land use,vegetation restoration and climate change to runoff and sediment discharge reduction.Meanwhile,based on multiple factors of land use,topography,soil type and vegetation coverage,new landscape metrics were constructed;then to quantify the relationship between landscape metrics and runoff and sediment yield by partial least squares method.For regional scale,attribution approaches were applied to quantify the contributions of climate change and vegetation restoration to runoff and sediment discharge reduction;then to quantify the relationship between landscape metrics and runoff,sediment transport coefficient of hydrological monitoring stations.The specific results and conclusions are as follows:（1）The runoff and erosion rates varied significantly among the conditions of horizontal,vertical,random and S-shaped flow paths（P<0.05）.The poorly-connected paths（horizontal and random）exhibited a stronger flow resistance and weaker sediment transport capacity than the well-connected paths（vertical and S-shaped）.The average flow rate and erosion rate of the vertical flow path were the highest,and the average rate of random flow path was the lowest.The runoff rates for the grassland and shrubland with coverage of 20-90%were decreased by 19.3-78.0%（P<0.05）,the sediment yield were reduced by 84.3-99.3%（P<0.05）,compared to the bare land.In addition,stems performed better than other components in runoff reduction,with mean relative contributions of 55.6%;biological crusts performed better than other components in soil erosion reduction,with mean relative contributions of 55.1%.The runoff reduction rate caused by the litter in the shrubland plots was approximately 49.9%,which was greater than that of the stems of shrubs and grasses,with relative contribution rates of approximately 3.3%and 4.8%,respectively.The litter had a significant impact on reducing sediment concentration,which was reduced by 94.8%,contributing to the sediment yield reduction by 22.2%.（2）The observed annual runoff at Ganguyi station exhibited a significant decreasing trend of 0.02 Gm³ yr^-1（P<0.01）and the annual sediment discharge decreased significantly of0.001 Gt yr^-1（P<0.01）during 1960-2015.Two relatively abrupt changes occurred around1979 and 1999 for all catchments based on the double mass curves.Accordingly,the study period was divided into three periods,which covered Period 1（1960-1979;P1）,Period 2（1980-1999;P2）,and Period 3（2000-2015;P3）.Compared to P1,the runoff showed a pronounced decrease of 32.4%in P2 and 49.2%in P3.The average annual runoff for the Yanhe catchments decreased by 8.8%and 33.0%in P2 and P3,respectively.The average sediment discharge decreased by 32.7%and 76.4%in P2 and P3,respectively.The water yield was approximately 35.6%lower in 2015 than in 1990,based on the SWAT simulations.We calculated that approximately 64.4%of the reduction in the total runoff were attributed to climate change from 1990 to 2015.In addition,the vegetation coverage reduced runoff by78.6%（0.33 Gt）during P3.（3）In our study,landscape metrics based on land cover type,topography,soil type and vegetation coverage is developed to evaluate the effect of watershed landscape factors on runoff and sediment yield.The partial least squares regression（PLSR）was used to analyze the relationship between landscape pattern indices and runoff,sediment yield.Landscape pattern is developing towards low connectivity,high fragmentation and high heterogeneity from 1990 to 2015 in the Yanhe watershed.For PLSR model between landscape pattern indices and water yield,sediment yield,the value of Q²>0.7,suggest that the performance of the model was satisfactory in this application.The Nash efficiency coefficient（NSE）shows that runoff and sediment yield prediction values in the Yanhe River Basin are in the credible range(NSE_runoff=0.83,NSE _{soil erosion}=0.42),and have a good correspondence with the measured values,which indicates the PLSR model can be used to simply predict basin runoff and sediment yield.（4）The results showed that the observed annual runoff at hydrological station（Huangfu,Wenjiachuan,Baijiachuan,Ganguyi）,which was distributed within the tributary of the Yellow River,exhibited a significant decreasing trend ranged from 0.02-0.15 Gm³ yr^-1（P<0.01）during 1960-2015.The observed annual sediment discharge exhibited a significant decreasing trend ranged from 0.001-0.031Gt yr^-1（P<0.01）.Compared to P1,the runoff showed a pronounced decrease of 32.4%in P2 and 49.2%in P3.The total average annual runoff for the four catchments decreased by 32.9%and 64.1%in P2 and P3,respectively.The average sediment discharge decreased by 50.6%and 87.2%in P2 and P3,respectively.The annual precipitation exhibited no significant decreasing trends from 1960 to 2015.Approximately 53.5%（3.81 Gm³）of the reduction in the total runoff during P2 and 6.8%（0.90 Gm³）during P3 were attributed to climate change;increasing vegetation coverage was the primary factor driving the reductions in runoff and sediment loads in P3 and accounted for 35.1%（0.26 Gm³）and 78.6%（0.33 Gt）of the total reduction.（5）The changed of landscape metrics in the Hekou–Longmen region from 1990 to 2014showed that the density of plaques became larger,the degree of heterogeneity and fragmentation increased,the diversity of landscape plaque shape decreased,and the degree of dispersion became higher.For PLSR model between landscape pattern indices for twenty hydrological station and runoff coefficient,the value of Q²>0.5,suggest that the performance of the model was satisfactory in this application.The value of NSE shown that the PLSR model has a good predictive effect on runoff coefficient in the basin and the model can be used to predict the runoff coefficients in the Hekou–Longmen region.